Historical research practices are being gradually transformed by the digitization of historic sources on the one hand, and the usage of digital methods and semantic technologies on the other. We present two digital initiatives of the Austrian National Library, ANNO and ONB Labs, which, as a digital hub of cultural heritage, enhances accessibility to and knowledge discovery within historical datasets considerably. To do so, ÖNB provides resources such raw data, metadata, or Linked Open Datasets which can be accessed as data dump or via a SPARQL API, allowing for live querying of RDF datasets. ONB Labs, additionally, offers services, tools, and APIs to enrich this data such as IIIF, Open Annotations, SACHA, or Jupyter Notebooks which allow to create and share documents including live code, visualizations and narrative texts. In this demopaper, we explore the question how digital methods and semantic technologies can be used in the context of historical research and illustrate their application through the example of a project investigating the cultural history of financial news in Vienna 1771-1914. We conclude with an outlook and call for action to encourage the use of ANNO and ONB Labs among digital historians.
Historical research practices are being gradually transformed by the digitization of
historic sources and the usage of digital methods and semantic technologies [
The advantages of integrating quantitative approaches against the background of
digitization are widely shared in historical studies and the possibilities of digitization are
increasingly being exploited. These initiatives can be summarized by the term Digital
History, a movement that understands itself as a branch of Digital Humanities with
overlaps on many levels [
With ANNO and ONB Labs, researchers are provided with a rich set of data, tools and services to explore historical data and contribute to digital history research. 3.1
In 2003, the Austrian National Library launched a digitization initiative focusing on
the transformation of historical newspaper and magazine collections into digital
copies. The result of this project is today available as ANNO (http://anno.onb.ac.at/anno),
a digital portal that makes the digitized records available free of charge and without
any time limit. The digitization of the historic material takes place continuously,
whereby the ranking of the titles to be digitized takes into account the interests of
researchers and prioritizes holdings with the highest intensity of use. ANNO pursues
the goal of offering collections that are as integrated and complete as possible by
collecting holdings from different archives and libraries. Currently, the ratio of fully
complete holdings is 22% (271 titles of 1,258), with the holdings being continually
supplemented and completed. As of April 2019, 1,258 titles were available in digital
format for the period from 1568 to 1947. The OCR-based full-text search covers more
than 90% of the digitized material with focus on the years 1689-1947. This
corresponds to 1,466,451 newspaper and magazine editions with nearly 20 million pages,
which are searchable at key-word level [
The full-text search is based on automatic indexing of source material by means of Optical Character Recognition (OCR). Due to still existing problems with the text recognition, ANNO inventories can also be very faulty in some texts. ANNO offers the possibility to specify the search with the help of further search options such as Boolean operators, wildcards, phrase search or distance search. The search results are displayed as a list of newspaper issues in which the given term occurs. The search results are sorted according to different criteria (relevance, date, title or the time of their inclusion into the index) and classified according to several categories (title, location, language, period, and topic). This display also provides information about the date of publication and the number of hits in the issue.
Thanks to its accessibility, scope, and quality, ANNO has become an important digital repository of historical newspapers. However, ANNO is still fraught with a few weaknesses, especially with respect to information extraction and semantic richness. The validity and usefulness of the analysis tools in form of filters are limited due to the lack of transparency regarding category definitions and search algorithms underlying the filter criteria. Some of these points can be compensated by ONB Labs which offer collaborative tools for enriching, processing and visualizing of datasets generated by ANNO. 3.2
The Austrian National Library conserves, archives and presents Austrian cultural heritage. The new methods to scan and provide printed materials led to an enormous increase in data, requiring new technologies to work with these masses of data, such as data mining, data linking, machine learning, artificial intelligence or algorithmic analysis. ONB Labs seeks to foster and support the application of these technologies.
Analogue to the conservation of memories in a consciousness system, a crucial part of long-term preservation of cultural heritage is its continuous usage and actualization. To ensure that the Austrian National Library establishes services and infrastructure dedicated to foster the work with its collections. Part of these services is the ONB Labs, a platform dedicated to support and inspire research, experimentation, and creative work with the Austrian National Library's digital collections and data. ONB Labs selects and presents openly licensed collections, tools to support working with them, and sketches to illustrate use cases. Data provided via ONB Labs is licensed under CC0 or published with a public domain mark. Via the IIIF API and interface SACHA (Simple Access to Cultural Heritage Assets) users can create collections including various data sets based on their specific needs and interests. These collections can be shared, cloned, exported to other tools, and downloaded.
One service implemented in ONB Labs is the provision of the Austrian National Library’s Linked Open Dataset. Linked Open Data refers to openly and freely available data online, identified via Uniform Resource Identifier (URI) and accessible via HTTP. This allows to link and send requests to various data resources. The Linked Open Data Set of the ONB Labs includes metadata of our historic newspapers, our historic postcards, and the catalogue of the Austrian National Library and can be accessed via ONB Labs as data dump and via SPARQL Lab, allowing for live querying of RDF datasets.
The focus group of ONB Labs are not solely Digital Humanists, but also digital artists, creatives and data scientists. Whereas this might be a diverse group of people, they are combined by working in fields where technology, research and work inform and guide each other reciprocally. Computational analysis is crucial within all of these areas. The provision of raw data, metadata, possibilities to enrich this data, OCR, tools and APIs tailored to the needs of programmers, are immanent to the work of these users and therefore the focal point of ONB Labs.
Sharing, commenting and forking is a reoccuring element in ONB Labs. Using Jupyter Notebooks allows to create and share documents including live code, visualizations, and narrative text. The ONB Labs team shares various notebooks and their documentation to allow users to use them as a basis for their own needs and further share them via the ONB Labs GitLab. 4
ANNO and some features of ONB Labs were key elements in a research project investigating the genesis of financial journalism and the popularization of financial news in Vienna 1771-1914. To answer the research questions, how financial journalism changed with respect to relevant newspapers, formats, and functions over time, and which public should be addressed with financial news in what form, we (1) searched ANNO for relevant key-phrases, (2) extracted the metadata associated with the search term (title, date of issue and number of hits) available on the website in HTML format using Web Scraper (webscraper.io) in a structured, statistically evaluable form usable for further analysis, (3) enriched the data with additional criteria such as main topic and target group, (4) visualized the data in form of interactive graphs using Python and Plotly (plot.ly), and (5) created collections of newspapers the visualizations are based upon using IIIF (iiif.io).
The dataset contains 4,028,974 hits spread across five terms: Börse* (1,685,441),
„Wiener Börse“ (235,597), Aktie* (1,158,329), Obligation* (594,106), and
Pfandbrief* (355,501). The query was made on 2019/02/19 and 2019/03/04. As newspapers
are digitized continuously, a query at this point in time would lead to different results.
For each term, more than 100 newspapers were identified, with more than 80% of all
hits being included in the top ten [
The intention of this demo-paper is to draw attention to the datasets, tools, and services provided by ÖNB to encourage collaborative practices in the enrichment and reutilization of historical datasets among digital historians. Components to annotate user-generated collections based on IIIF and Open Annotations are continuously implemented and in the near future annotations will be sharable – both publicly and privately. We are especially eager to explore the intersections of historical data, linked data, and visualization, and provide support for collaborations via the services of ONB Labs. We invite all interested parties get in touch with OBN Labs and share their ideas.